Abstract Because the ovarian response to FSH stimulation in assisted reproduction is variable, ranging from hyporesponse to hyperresponse, with the possible complication of ovarian hyperstimulation, it would be of great benefit to predict the response of the patients to FSH. To date, no clear-cut predictors of ovarian responsiveness to FSH have been identified. In this study, we investigated the role of two distinct FSH receptor (FSHR) variants, Thr307/Asn680 and Ala307/Ser680, in the response to FSH in women undergoing controlled ovarian stimulation. The FSHR polymorphism at position 680 was analyzed by restriction-fragment-length polymorphism in 161 ovulatory women below the age of 40 yr. With reference to the couple, infertility has been diagnosed as being attributable to male causes (76%), tubal factor (11%), or both (13%). The distribution was 29% for the Asn/Asn, 45% for the Asn/Ser, and 26% for the Ser/Ser FSHR variant. Peak estradiol levels, number of preovulatory follicles, and number of retrieved oocytes were similar in the 3 groups. However, basal FSH levels were significantly different among the 3 groups (6.4 ± 0.4 IU/L, 7.9 ± 0.3 IU/L, and 8.3 ± 0.6 IU/L for the Asn/Asn, Asn/Ser, and Ser/Ser groups, respectively, P < 0.01). The number of FSH ampoules required for successful stimulation was significantly different among the 3 groups (31.8 ± 2.4, 40.7 ± 2.3, and 46.8 ± 5.0 for the Asn/Asn, Asn/Ser, and Ser/Ser groups, respectively, P < 0.05). According to multiple linear regression analysis, the number of ampoules needed could be predicted from a linear combination of both the type of polymorphism and basal FSH levels (P < 0.001). These clinical findings demonstrate that the ovarian response to FSH stimulation depends on the FSHR genotype.